mullikine

This is a library that lets you interface with language models through programming. It separates the peas and carrots of real and imaginary, through lisp function and macro syntax.

Sure thing! Add me on linkedin and flick me a message. Busy until the 25th though

Thank you! I have made the corrections. It's still fresh!

    - Chatbots for whatever you are looking at
    - Complex terminal to add Codex/GPT to any terminal program
    - Paracosm: Mind mapping with GPT-3
    - LookingGlass: A browser for the imaginary web
      - Dream up websites from URLs or missing pages

All integrated!

You can use Complex Term without emacs. Just prefix `ct` to any shell command.

For example:

    ct vim my-file.txt

Run the web browser like this:

    lg "https://semiosis.github.io/ii/"

Just add the scripts directory to your path.

Please try out Pen: It's ready!

Language models for the terminal. https://semiosis.github.io/cterm/

Coherent, relevant chatbots for anything you're looking at. https://semiosis.github.io/posts/multi-part-prompts/

A browser for the imaginary web. https://semiosis.github.io/looking-glass/

Imaginary interpreters. https://semiosis.github.io/ii/

Mind mapping with chatbots, auto suggested topic generation, etc. https://semiosis.github.io/paracosm/

This is an LSP server for OpenAI's GPT-3/Codex or any language model, using emacs. It gives documentation and refactoring to any language including world languages and fictional ones. As an LSP server, it works for VSCode too.

Firstly, what I'm saying does not rest on blockchain being the source of truth. It could, for example, be another language model from a corporation that you trust. Secondly, if you can't see the value of a blockchain or decentralised cryptography as a way to deal with the issues of consensus, then that's not my problem.

This is just wrong. Imaginary computing (based on language models) and blockchain or other source of truth is the future. As hermit programmers we can rely on AI to make a prediction as to when a task will be done, generate documentation for us, generate tests https://mullikine.github.io/posts/imaginary-reflection/, generate, etc. Any deployment system that doesn't use language models will be fragile as the time to onramp new developers will become exponentially more difficult over time. Hermit programmers are OK if they lean heavily on AI and blockchain. This article is a case for sell-outs, and pessimists

This is an example of an imaginary web browser to surf the imaginary internet, using OpenAI's Codex. It's built into Pen.el and is completely free as in freedom.

https://news.ycombinator.com/item?id=28489942

I made a post on HN but didn't get enough hits, so crossposting to relevant HN posts such as this.

https://semiosis.github.io/posts/the-imaginary-web-with-code...

- Visit any website you can imagine, even the ones that are not real!

- Edit and re-imagine as you go see alternative website realities – change the sentiment of the author!

- Peer into the future – read about GPT-5!

- Generate relevant URLs (often real, sometimes imaginary) from any text selection

- Read an article on anything from your favourite blogger.

- … This is the future of the web.

Visit any website you can imagine, even the ones that are not real!

Edit and re-imagine as you go see alternative website realities – change the sentiment of the author!

Peer into the future – read about GPT-5!

Generate relevant URLs (often real, sometimes imaginary) from any text selection

Read an article on anything from your favourite blogger.

… This is the future of the web.

This is an example of an imaginary web browser to surf the imaginary internet, using OpenAI's Codex. It's built into Pen.el and is completely free as in freedom.

More could be done to fund research that showcases the importance and value of transformative models, so that countermeasures to existential threats have time to be developed. So long as there is a practical strategy for mitigating the philosophical problems such as software licenses and risk-managing existential threats to public safety, I think OpenAI deserves the liberty to take a profit. But exclusively licensing such models to corporations is very harmful to liberty and continual exaccerbation of the imbalance of economic power will lead to humanicide. All people need to benefit from this technology, or the people that are ignored will suffer from it.

But this problem is solved using GitHub's CodeQL for searching and filtering generated code. By combining Copilot with GitHub Semantic and GitHub CodeQL, you have a means of writing and generating the code you want in a secure way. This means that you no longer need the original source code that was used to train Codex. Training Codex and selling as a product in the form of Copilot steals the essence of the original source code used to train it, to build the future of programming, while paying nothing back to the original authors. Even Elon Musk was opposed to OpenAI exclusively licensing to Microsoft GPT.

https://edition.cnn.com/2020/09/27/tech/elon-musk-tesla-bill...

It's so transformative that people may allow it to circumvent licenses.

These language models are not being utilised very well by tools such as copilot because they are not mapping very many functions from the editor to language model. The more functions you map, the more you get out of it. If Copilot's workings were completely open and configurable, you would find that people collectively can work together to map many functions to the language model. They are capable of far greater wonders with deep integration and collaboration. I have tried to demonstrate this with emacs.

https://news.ycombinator.com/item?id=28045894

I wholeheartedly agree. It's just obvious that this is harvesting work done by the free software community. It will be very obvious in retrospect, but hard to see now. If you consider that people will code more abstractly from here on, using AI code generation and understanding, automating workflows, the real value is in the way software is _used_, over the original source code. This is what GitHub has stolen (the representation of software as defined by its usage). Like a function can be defined by a formula or by its domain and range, so does software have multiple reprsentations. The representation of how software is used as being just as important as how it was written will become obvious in the future. GitHub should start by serving a model trained on ONLY free software because right now, in order to remain pure and keep separate from SAAS and copilot, we are losing productivity. It's not fair on open source!

Well, almost Libre! The SAAS needs to be replaced. But at least it's using a libre model from EleutherAI.

Pen.el now can take advantage of GPT-J-6B via AIX. That means there is a completely free (libre) alternative to VSCode and Copilot.

https://www.reddit.com/r/emacs/comments/owwcwv/gptj_support_...

Get yourself a key if you'd like to try it out.

https://apps.aixsolutionsgroup.com/

You may make composite 'prompt functions' that combine both language models, and also fine-tuned OpenAI models and use them to extend emacs in all directions!

Prompt chaining: Linking prompts together usefully in emacs

https://mullikine.github.io/posts/using-emacs-prompt-functio...

Aside from being a local celebrity, Alex services pianos around Dunedin. Who better to ask to tune my piano than Alex. And he blessed me with a tour of his piano shop and a concert on the long piano. He is the coolest guy ever. I asked him about the possibility of making a piano which two people could play -- in the same key and he entertained the idea. He is so passionate. Such a cool person!

A prompt for touching up those commit messages would be awesome!

This is an update to my previous submission a week ago.

- Easy docker install and run - one-liner - A tutorial - Rewritten most of the code to be - faster, - more featureful, - more reliable, and - much closer to pure elisp

pen.el is a GPT-3 editor based on emacs and an alternative to Copilot and is configurable to connect other completion engines. It's also an emacs lisp package that can be installed into your existing environment.

I am working on the GPT-j completer currently. Google has kindly sponsored a free TPU. Thank you Google!

I'm camped out in McDonalds using the free WiFi to upload the image. Thanks for sponsoring some free WiFi too, McDonalds!

Discard the docker image after use as it saves the OpenAI API key.

  docker run --rm -ti --entrypoint= semiosis/pen.el:latest ./run.sh

Please give it a try and submit bug reports. P2P prompt sharing is on its way.

Also mark, I would like to use your completion code as part of a back-end general completion script for pen.el, if that's ok, as I think building this in clojure is 'the way'.

https://github.com/semiosis/pen.el

https://github.com/semiosis/lm-complete

Paid, thank you!

Love your work Mark. I have been digging into your book. I'm looking to incorporate Datomic into my project to store generations from GPT-3. Will be taking quite a few pages from your book. Thanks! I'm getting around to paying for your book, sorry I haven't done it yet. https://semiosis.github.io/posts/representing-multiverses-wi...

Public facing open-source code & media is going to be learned by language models because they're exposed to them. That's the simple truth. Nothing can stop that, not unless all public repos are made private. Everyone has access to the ability to create their own GPT, thanks to open-source. OpenAI is not actually very far ahead of open source anymore.

The US seems well enough informed. As mentioned in the following report "AI tools are diffusing broadly and rapidly" and "AI is the quintessential “dual use” technology—it can be used for civilian and military purposes.".

https://www.nscai.gov/wp-content/uploads/2021/03/Full-Report...

I'm fully expecting that if I begin a story and put it on my blog or on github, and if I go away for a couple years, I'll see it completed for me when I return. I can use foresight to my advantage or I can pretend like it's still the 1990s as if placing some text at the top of the code I exposed publicly is going to prevent people from training on it.

One thing for sure though, I don't think a large company such as Microsoft should be profiting from training their language model on open-source code.

The best way to release Copilot in my opinion would be to make the entire thing open source and have separate models, even a private paid-for model so long it's trained on their own code.

An open source model trained on code for specific licenses sounds fine, but then the model should also follow that same license as the code it was trained on.

There's just something deeply unsettling about having a computer complete your thoughts for you without being able to question how or why.

The cat (this technology) is coming out of the bag one way or another. It's just too useful. Where is it written that inspiring a language model with data is not just as infringing as copy and paste? It's a very grey line. Public facing open-source code & media is going to be learned by language models because they're exposed to them. I'm fully expecting that if I begin a story and put it on my blog or on github, and if I go away for 5 years, I'll see it completed for me when I return.

I'm trying to get it going in a docker container asap.. I was slaving away on that today